[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M7AD06EWF2TM1D2HPAQD8KPP":3,"skill-same:01M7AD06EWF2TM1D2HPAQD8KPP":50},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":15,"categories":16,"keywords":17,"publishedAt":18,"linkOnly":19,"markdownContent":20,"markdownBody":21,"frontMatter":22,"refName":23,"commitSha":24,"skillDirectory":25,"forksCount":26,"repositoryHtmlUrl":27,"repositoryHomepageUrl":6,"ownerHtmlUrl":28,"ownerBlogUrl":6,"ownerDescription":6,"updatedAt":29,"install":30,"resources":43},"01M7AD06EWF2TM1D2HPAQD8KPP","stay-within-limits",null,"Use when long-running or parallel agent work must respect 5-hour and weekly usage limits by checking usage between waves, pausing near the cap, and resuming only when the window is clear.","JesterWizard\u002FSkillSystem_FE7","JesterWizard","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F164426396?v=4","01M7ACZD32AVSPDR22QKQR5HNY","01M7ACZD478KR9RA07QGPHE24H",4,"CC0-1.0","https:\u002F\u002Fgithub.com\u002FJesterWizard\u002FSkillSystem_FE7\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fstay-within-limits\u002FSKILL.md",[],[],"2026-11-15 14:19:17",false,"---\nname: stay-within-limits\ndescription: Use when long-running or parallel agent work must respect 5-hour and weekly usage limits by checking usage between waves, pausing near the cap, and resuming only when the window is clear.\n---\n\n# Stay Within Limits\n\nKeep long-running agent work inside the current 5-hour and weekly usage windows.\nCheck usage before launching substantial work and between waves of parallel\nsubagents. If an active 5-hour or weekly limit is at or above 95%, pause new\nwork until the window is clear enough to continue safely.\n\n## Core Loop\n\n1. Run a bounded wave of work. Default to at most 3 parallel subagents unless\n   the user or host gives a different throttle.\n2. Wait for the wave to finish. Do not interrupt in-flight subagents just to\n   save budget; that usually loses work.\n3. Check current 5-hour and weekly usage with the host's usage\u002Fbudget tool.\n4. If either window is at or above 95%, stop launching work and schedule a\n   self-contained resume when the relevant window should clear.\n5. On resume, re-check the real window or block before continuing. Do not trust\n   elapsed wall-clock time alone.\n\n## Usage Signals\n\nPrefer a first-party host usage tool when available. In Claude Code, use:\n\n```sh\nnpx -y ccusage@latest blocks --active --json\n```\n\nUse the JSON to identify the active block start, current cost or percentage, and\ntime remaining. On wake, compare the active block start timestamp with the\nprevious one; a new timestamp is stronger evidence than \"enough time passed.\"\n\nIf the tool reports cost instead of a direct percentage, convert through the\ncurrent account limit when known. For Claude Max-style 5-hour blocks, some users\nprefer an earlier caution threshold around $500-550; treat that as a\nuser-configured guardrail, not a universal rule. The default stop rule is still\n95% of the active 5-hour or weekly limit.\n\n## Pausing And Resuming\n\nWhen a wake\u002Fresume tool is available, schedule a wakeup for:\n\n```txt\nmin(3600, secondsUntilWindowClears)\n```\n\nIf the runtime clamps wake delays to 60-3600 seconds, chain wakeups for longer\nwaits. Each wakeup should re-check usage, reschedule if still over budget, and\ncontinue only when the window is safely below the threshold.\n\nMake wake prompts self-contained. Include:\n\n- The remaining plan.\n- The check-then-reschedule rule.\n- The 95% threshold and wave throttle.\n- The exact usage command or host usage tool to run.\n- The previous block\u002Fwindow identifier when available.\n- The next verification steps.\n- The next wave's handoff packets, including scope, verification commands, and\n  stop conditions, if delegation will resume.\n\n## Choosing The Wait Mechanism\n\n- Use a wake\u002Fresume tool when the agent needs instructions attached to the\n  future resume.\n- Use a background sleep or watcher for fixed timers and things a process can\n  observe directly.\n- Use cron or recurring schedules only for recurring fresh-session work.\n\nAvoid short-interval polling for things the host will notify you about, such as\nbackground task or subagent completion. For budget pauses, a prompt-cache miss\nafter a long sleep is acceptable; preserving the limit matters more.\n\n## Reporting\n\nIf you pause, tell the user which window is over threshold, the observed usage,\nwhen you scheduled or expect the next check, and what work remains. Keep enough\nstate in the wake prompt that the next turn can resume without relying on\nconversation momentum.","# Stay Within Limits\n\nKeep long-running agent work inside the current 5-hour and weekly usage windows.\nCheck usage before launching substantial work and between waves of parallel\nsubagents. If an active 5-hour or weekly limit is at or above 95%, pause new\nwork until the window is clear enough to continue safely.\n\n## Core Loop\n\n1. Run a bounded wave of work. Default to at most 3 parallel subagents unless\n   the user or host gives a different throttle.\n2. Wait for the wave to finish. Do not interrupt in-flight subagents just to\n   save budget; that usually loses work.\n3. Check current 5-hour and weekly usage with the host's usage\u002Fbudget tool.\n4. If either window is at or above 95%, stop launching work and schedule a\n   self-contained resume when the relevant window should clear.\n5. On resume, re-check the real window or block before continuing. Do not trust\n   elapsed wall-clock time alone.\n\n## Usage Signals\n\nPrefer a first-party host usage tool when available. In Claude Code, use:\n\n```sh\nnpx -y ccusage@latest blocks --active --json\n```\n\nUse the JSON to identify the active block start, current cost or percentage, and\ntime remaining. On wake, compare the active block start timestamp with the\nprevious one; a new timestamp is stronger evidence than \"enough time passed.\"\n\nIf the tool reports cost instead of a direct percentage, convert through the\ncurrent account limit when known. For Claude Max-style 5-hour blocks, some users\nprefer an earlier caution threshold around $500-550; treat that as a\nuser-configured guardrail, not a universal rule. The default stop rule is still\n95% of the active 5-hour or weekly limit.\n\n## Pausing And Resuming\n\nWhen a wake\u002Fresume tool is available, schedule a wakeup for:\n\n```txt\nmin(3600, secondsUntilWindowClears)\n```\n\nIf the runtime clamps wake delays to 60-3600 seconds, chain wakeups for longer\nwaits. Each wakeup should re-check usage, reschedule if still over budget, and\ncontinue only when the window is safely below the threshold.\n\nMake wake prompts self-contained. Include:\n\n- The remaining plan.\n- The check-then-reschedule rule.\n- The 95% threshold and wave throttle.\n- The exact usage command or host usage tool to run.\n- The previous block\u002Fwindow identifier when available.\n- The next verification steps.\n- The next wave's handoff packets, including scope, verification commands, and\n  stop conditions, if delegation will resume.\n\n## Choosing The Wait Mechanism\n\n- Use a wake\u002Fresume tool when the agent needs instructions attached to the\n  future resume.\n- Use a background sleep or watcher for fixed timers and things a process can\n  observe directly.\n- Use cron or recurring schedules only for recurring fresh-session work.\n\nAvoid short-interval polling for things the host will notify you about, such as\nbackground task or subagent completion. For budget pauses, a prompt-cache miss\nafter a long sleep is acceptable; preserving the limit matters more.\n\n## Reporting\n\nIf you pause, tell the user which window is over threshold, the observed usage,\nwhen you scheduled or expect the next check, and what work remains. Keep enough\nstate in the wake prompt that the next turn can resume without relying on\nconversation momentum.",{"name":5,"description":7},"main","b0f8a46e7eb03e690c3965dbe08b10b1bada7364",".claude\u002Fskills\u002Fstay-within-limits",0,"https:\u002F\u002Fgithub.com\u002FJesterWizard\u002FSkillSystem_FE7","https:\u002F\u002Fgithub.com\u002FJesterWizard","2026-09-07 02:17:07",{"source":8,"skillName":5,"commands":31,"prompt":41,"zipAvailable":42},[32,35,38],{"tool":33,"command":34},"npx","npx skills add JesterWizard\u002FSkillSystem_FE7 --skill stay-within-limits",{"tool":36,"command":37},"bunx","bunx skills add JesterWizard\u002FSkillSystem_FE7 --skill stay-within-limits",{"tool":39,"command":40},"pnpm","pnpm dlx skills add JesterWizard\u002FSkillSystem_FE7 --skill stay-within-limits","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002FJesterWizard\u002FSkillSystem_FE7\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fstay-within-limits\u002FSKILL.md",true,[44],{"path":45,"resourceType":46,"resourceTypeDesc":47,"sizeBytes":48,"sourceHtmlUrl":49},"README.md","REFERENCE","参考文档",2422,"https:\u002F\u002Fgithub.com\u002FJesterWizard\u002FSkillSystem_FE7\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fstay-within-limits\u002FREADME.md",[51,58,65,72,79,84],{"skillNo":52,"name":53,"nameZh":6,"description":54,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":55,"categories":56,"keywords":57,"publishedAt":18},"01M7AD0HNWAZ0QK1ZKDM7WTRVR","implement-skill","Resolves a named FE7 skill to its file pack and edits only those paths. Use when the user wants to implement, enable, port, wire, fix, add, install, or debug a skill (Nullify, Provoke, Adept, etc.).","https:\u002F\u002Fgithub.com\u002FJesterWizard\u002FSkillSystem_FE7\u002Fblob\u002Fmain\u002F.cursor\u002Fskills\u002Fimplement-skill\u002FSKILL.md",[],[],{"skillNo":59,"name":60,"nameZh":6,"description":61,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":62,"categories":63,"keywords":64,"publishedAt":18},"01M7AD0DM489QEVQ01M9GNMSXE","documentation-style-docs","Use when creating or updating project documentation in the same architecture and writing style as FogStages.md: title + visual block, index, introduction, implementation plan, code locations table, TODO, and limitations\u002Fbugs. Trigger on requests like create docs, write documentation, add design document, feature spec, system write-up, or architecture note.","https:\u002F\u002Fgithub.com\u002FJesterWizard\u002FSkillSystem_FE7\u002Fblob\u002Fmain\u002F.cursor\u002Fskills\u002Fdocumentation-style-docs\u002FSKILL.md",[],[],{"skillNo":66,"name":67,"nameZh":6,"description":68,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":69,"categories":70,"keywords":71,"publishedAt":18},"01M7AD09G3JN4Z7DRG7QY84PA2","caveman","Ultra-compressed communication AND thinking. Always use extreme (ultra) caveman to minimize tokens in both reasoning and replies. Cuts output ~65% and thinking tokens by skipping recap, option-lists, and filler thought. Use when user says \"caveman mode\", \"talk like caveman\", \"use caveman\", \"less tokens\", \"be brief\", \"extreme caveman\", or invokes \u002Fcaveman. Also auto-triggers when token efficiency is requested.","https:\u002F\u002Fgithub.com\u002FJesterWizard\u002FSkillSystem_FE7\u002Fblob\u002Fmain\u002F.cursor\u002Fskills\u002Fcaveman\u002FSKILL.md",[],[],{"skillNo":73,"name":74,"nameZh":6,"description":75,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":76,"categories":77,"keywords":78,"publishedAt":18},"01M7AD02NZX9W1V2XVTKWARR6M","quick-recap","Use when adding or following the red\u002Fyellow\u002Fgreen final status block convention for agent responses, especially by installing managed AGENTS.md or CLAUDE.md instructions.","https:\u002F\u002Fgithub.com\u002FJesterWizard\u002FSkillSystem_FE7\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fquick-recap\u002FSKILL.md",[],[],{"skillNo":80,"name":53,"nameZh":6,"description":54,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":81,"categories":82,"keywords":83,"publishedAt":18},"01M7ACZZTMSRS96D0H2FNM35T0","https:\u002F\u002Fgithub.com\u002FJesterWizard\u002FSkillSystem_FE7\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fimplement-skill\u002FSKILL.md",[],[],{"skillNo":85,"name":86,"nameZh":6,"description":87,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":6,"repositoryNo":12,"starsCount":13,"licenseSpdxId":14,"licenseFileUrl":6,"sourceHtmlUrl":88,"categories":89,"keywords":90,"publishedAt":18},"01M7ACZWMWHQMFG0D82K2AVZPZ","efficient-frontier","Apply the same orchestration as `\u002Fefficient-fable` to any high-cost frontier model: delegate research, coding, and testing to cheaper subagents while keeping planning, synthesis, and final review with the expensive model.","https:\u002F\u002Fgithub.com\u002FJesterWizard\u002FSkillSystem_FE7\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fefficient-frontier\u002FSKILL.md",[],[]]